1# @(#)README.ATAPI 1.7 20/11/21 Copyright 1997-2004 J. Schilling 2 3People (with a Linux only background) often ask me why do you depend on 4"ATAPI-SCSI emulation", why don't you support generic IDE? 5 6First a note: there is no SCSI emulation on Linux, but there has been 7some kind of SCSI integration. To explain the difference between SCSI emulation 8and SCSI integration let us look at an Operating system that implements both, 9Windows NT: 10 11Win NT implements SCSI integration, if you open "\\\\.\\SCSI%d:", BusNo, 12you will be able to send generic SCSI commands to any SCSI speaking drive 13on that bus. ATAPI drives just show up on a specific Busnumber, SCSI drives 14that use a 50 or 68 pin connector show up on different Bus numbers. 15Note that Microsoft obviously did copy my ideas I did implement in August 1986 16for SunOS-3.0 :-) 17 18Win NT implements SCSI emulation by not only showing drives that natively 19talk SCSI on the ATA/IDE cable, if you send generic SCSI commands via ioctl's 20to "\\\\.\\SCSI%d:", BusNo. Win NT also shows you plain ATA drives that do not 21understand SCSI commands in firmware. Win NT allows you to send SCSI 22commands to the kernel and the kernel translates the SCSI commands to ATA 23commands with the same meaning, so Win NT emulates SCSI for ATA drives. 24 25Linux does not emulate SCSI command transport for vanilla ATA drives, so there 26is definitely no SCSI emulation in Linux. 27 28 29 You need to believe me: There is no single IDE burner out! 30 Even a CD-ROM cannot be used decently if you use only IDE/ATA commands. 31 Opening/closing the door, playing audio and similar things 32 cannot be done using vanilla IDE commands - you will need SCSI commands 33 to do this. But how do we do this with a drive that uses an IDE 34 interface? 35 36 ATAPI stands for ATA Packet Interface 37 38 The ATAPI standard describes method of sending SCSI commands over IDE 39 transport with some small limitations to the "real" SCSI standard. 40 SCSI commands are send via IDE transport using the 'ATA packet' 41 command. There is no SCSI emulation - ATAPI drives include native 42 SCSI command support. For this reason, sending SCSI commands to ATAPI 43 drives is the native method of supporting ATAPI devices. Just imagine 44 that IDE is one of many SCSI low level transport mechanisms. 45 46 This is a list of some known SCSI transports: 47 48 - Good old Parallel SCSI 50/68 pin (what most people call SCSI) 49 - SCSI over fiber optics (e.g. FACL - there are others too) 50 - SCSI over a copper variant of FCAL (used in modern servers) 51 - SCSI over IEEE 1394 (Fire Wire) 52 - SCSI over USB 53 - SCSI over IDE (ATAPI) 54 55 As you now see, the use of the naming convention "ATAPI-SCSI emulation" 56 is a little bit misleading. It should rather be called: 57 "IDE-SCSI host adapter emulation" 58 59Some naming explanations: 60 61 ATA Attachment Adapter 62 IDE Integrated Drive Electronics (A Drive that includes ATA) 63 ATAPI ATA Packet Interface 64 65When cdrecord has problems with ATAPI drives on Linux this usually is a Linux 66kernel problem. The Linux kernel maintainers unfortunately refuse to correct 67their current IDE driver system setup which does not support ATAPI by default. 68ATAPI _is_ SCSI over IDE transport. It is hard to understand why Linux still 69uses a default driver setup that is designed for IDE CD-ROM drives made 70before 1994 (using a IDE compat mode that only allows to use the drive 71read-only) and does not handle to send SCSI commands to ATAPI drives by 72default. This makes it hard for people who just started with Linux to do 73CD-writing on Linux if they own an ATAPI drive. Both Linus Torvalds and 74Alan Cox admit that they don't own a CD/DVD writer, how should they know about 75the problems? 76 77There are bugs with the DMA implementation that are known for many years 78but they don't get fixed. 79 80/*--------------------------------------------------------------------------*/ 81Which Operating systems support ATAPI 82 83- AIX: Status unknown! Please report your experience... 84 85- Apple Mac OS X (Darwin): Supported 86 87- BeOS (libscg maps ATAPI to SCSI bus # >= 8 88 89- BSD/OS: Status unknown! Please report your experience... 90 91- FreeBSD: 92 - YES for the latest default kernel. 93 It includes finally ATAPI-Cam 94 95 - NO for the older kernels. 96 Yes, if you install a kernel patch from 97 Thomas Quinot <thomas@cuivre.fr.eu.org> 98 See http://www.cuivre.fr.eu.org/~thomas/atapicam/ 99 and README.FreeBSD 100 101- HP-UX: It looks like ATAPI does not work correctly due to kernel bugs. 102 103 New information: 104 HP supports a HP A7853A B/C class machine (s700_800) with HP-UX-11.x 105 You need to install a patch: 106 107 Patch Name: PHKL_27224 Patch Description: s700_800 11.00 IDE/ATAPI 108 cumulative patch 109 110 111- Linux (unfortunately not in the default configuration) 112 113 - It works more or less if you include ide-scsi 114 115 - Linux-2.4.xx includes a CDROM Packet interface in the 116 IDE CD driver. For this driver libscg now includes 117 support in pre-alpha status. Use cdrecord dev=ATAPI -scanbus 118 to check for drives and e.g. cdrecord dev=ATAPI:0,0 .... 119 for writing. Note that this interface is not integrated into 120 the standard libscg device naming scheme. Support for 121 this interface has been included because it is the only 122 way to use a PCCARD/PCMCIA writer - trying to use ide-scsi 123 on a PCATA interface will cause a Linux kernel panic 124 or will block all ATAPI drives. 125 126 - Starting with Linux-2.5.45, there is a new experimental 127 ATAPI interface initiated by Linus Torvalds. Unfortunately, 128 this interface does not fit well into the rest of the Linux 129 SCSI kernel transport naming scheme. Cdrecord allows to 130 use this interface by calling e.g. cdrecord dev=ATA:1,0,0 ... 131 132 All Linux ATAPI transport implementations do not support DMA. 133 Current exceptions are: 134 135 - ide-scsi with block size 2048 and if DMA has been enabled 136 137 - The new experimental ATAPI interface starting with Linux-2.5.45 138 allows DMA if DMA has been enabled and the sector size is a 139 multiple of 4. This allows to use DMA for audio CDs and 140 when writing any type of CD in RAW mode. 141 142 Note that is a bad idea to first implement a new kernel 143 interface that also implements the named DMA implementation 144 bugs and later fix the DMA bug _only_ for this new 145 interface. It looks like the Linux kernel folks are not 146 very cooperative :-( 147 148 RAW mode is needed for many new and cheap drives that have bugs when 149 writing in cooked mode. If there is no DMA, you cannot write faster 150 than approx 16x. 151 152 153- NetBSD (releases 1.3 and newer) 154 155- NeXT: Status unknown! Please report your experience... 156 157- OpenBSD: (release 2.6 and newer) 158 159- OS/2 (you need to fetch and install ATAPI support first) 160 see: http://www.leo.org/pub/comp/os/os2/leo/drivers/dasd/daniatapi.zip/ 161 162- OSF-1 / True64 Status unknown! Please report your experience... 163 164- SCO-OpenServer: Supported with 5.0.6 and non-public patch or with 165 5.0.7. I don't know whether you need a patch for 5.0.7 166 167- SCO-UnixWare: partial support with UnixWare 7.1.3 - some SCSI commands 168 that are needed for cdda2wav and DVD writing are blocked. 169 7.1.4 will have full ATAPI support. 170 171- SGI/IRIX: Status unknown! Please report your experience... 172 173- Solaris (you may need to use the USCSI transport interface to address 174 ATAPI if the IDE hostadapter idriver implementation does not follow 175 Sun's internal standards). 176 177 ATAPI works fine on Solaris 7 sparc and on Solaris 7/8 intel. 178 179 On Solaris 8 (intel) and newer, the ATAPI/SCSI subsystem is integrated 180 correctly according to Sun's SCSA white paper, so the 'scg' driver works. 181 This is not true for Solaris on sparc where the ATAPI driver do not conform 182 to Sun's internal structuring rules. You need to use the USCSI interface 183 on for ATAPI drives on Solaris sparc for this reason. 184 185 Solaris 8 sparc has a ATA DMA bug that prevents cdrecord from working at all. 186 There is a fix from Sun available: the patch 108974-16 187 Solaris 9 sparc works again, it has the fix for the ATA DMA bug included. 188 189 Newer versions of Solaris 9 disable DMA for CD-ROM drives on IDE. 190 Read README.solaris-x86-ATAPI-DMA to learn how this may be circumvented. 191 192- VMS: works on recent versions! 193 194- Win32 using a recent ASPI Layer supports ATAPI 195 You nay need to exclude mini port drivers (see README.win32). 196 197 Newer cdrecord versions also support the SPTI (SCSI Pass through ioctl). 198 Libscg uses SPTI by default if you are running NT-5.x or newer and are 199 administrator. With NT-4.x it may be possible to run cdrecord dev=SPTI:1,0.0 ... 200 But there are reports for blue screens (kernel crashes). 201 202- DOS DOS-7 from win98 includes a ATAPI aware aspi 203 For other versions have a look at README.msdos and use e.g. "oakaspi". 204 205/*--------------------------------------------------------------------------*/ 206General hints: 207 208*********************** 209NOTE: IDE/ATAPI doesn't have disconnect/reconnect! you cannot expect the needed 210 performance for CD-writing if you connect source and destination drive 211 to the same IDE cable. 212*********************** 213 214If you never like to directly write a CD from CD-ROM source, this configuration 215seems to be the best: 216 217IDE 0 MSTR -> HD1 218IDE 0 SLAV -> HD2 219 220IDE 1 MSTR -> CD-writer 221IDE 1 SLAV -> CD-ROM 222 223If you like to write from both HD source and CD-ROM source, you should have 224the following configuration: 225 226IDE 0 MSTR -> HD1 (does _not_ hold CD mastering data) 227IDE 0 SLAV -> CD-Writer 228 229IDE 1 MSTR -> HD2 (holds CD mastering data) 230IDE 1 SLAV -> CD-ROM 231 232If cou cannot set up a decent cabling (e.g. because you use a notebook) 233you may try to use cdrecord -immed ... 234It runs slow commands in quick (immediate) return background mode and 235tries to wait between the write commands to allow to free the IDE cable 236so the cdrecord read process may fill the FIFO from the other drive 237on the same IDE cable. 238 239/*--------------------------------------------------------------------------*/ 240The rest of this file is only valid for Linux! 241 242This was taken out of mails from From: Dave Cohen <dcohen@richmond.infi.net> 243and From: Gadi Oxman <gadio@netvision.net.il> 244(slightly modified marked ***JS *** except typo corrects) 245 246As all actual Linux versions have ATAPI support for cdrecord, 247I removed the patch section. If you are running a Linux version 248that does not support ATAPI<->SCSI command transport, please upgrade. 249 250The basic driver design in Solaris would also allow to use ATAPI 251drives but unfortunately, Sun made a mistake in the mid-level design. 252If you want to use ATAPI drives with Solaris, ask Sun why they don't 253support SCSI passthrough to IDE although they are using a common driver 254concept. 255 256Please use cdrecord-1.6 final or later (if available), it includes the 257modifications needed for ATAPI drives and is still working with other 258SCSI drives. Older revisions of cdrecord do not support ATAPI drives. 259 260If you are using Linux Kernel version prior to 2.1.73 or prior to 2612.0.35, please upgrade before you try to compile and use cdrecord. 262 263In any case, you need to configure a kernel with ATAPI/SCSI hostadapter 264emulation. Read carefully the following instructions: 265 266In any case, you need to disable generic IDE/ATAPI CDROM support in 267order to make ATAPI SCSI emulation working. 268 269Many people ask why I use ATAPI-SCSI emulation. 270 271 The use of the naming convention "ATAPI-SCSI emulation" is a 272 little bit misleading. It should rather be called: 273 "SCSI host adapter emulation" 274 275 The ATAPI standard describes method of sending SCSI commands over IDE 276 with some small limitations to the "real" SCSI standard. 277 For this reason ATAPI-SCSI emulation is the native method of 278 supporting ATAPI devices. 279 280If you have problems to talk to the device when it is jumpered as "slave" 281try to use it as "master". If you connect a hard disk to the same IDE 282cable as the CD writer or if you try to read/write data from another drive 283that is connected to the same IDE cable as the CD writer you may get 284problems too. 285 286NOTICE: 287 288With the newer 2.1.x or 2.2.x kernels it seems to be possible to run 289SCSI/ATAPI hostadapter emulation and generic IDE at the same time by 290selectively telling the kernel what to use for which drive. However, 291this would not be needed if the Linux SCSI CD-ROM driver would be more 292up to date and supports standard conforming drives. 293 294J�rg Schilling <joerg@schily.net> 295 296-------------------------------------------------- 297Here is a hint from Alan Brown <alanb@manawatu.gen.nz>: 298 299To allow ATAPI cd and ide-scsi support on the same machine, add 300`hd<x>=ide-scsi` to the lilo.conf append entry, or use 301`hd<x>=ide-scsi` at the bootup lilo prompt. 302 303I have my HP-7200 RW drive as the primary drive on the second IDE 304bus, so the statement used is "hdc=ide-scsi" 305 306-------------------------------------------------- 307 308Hope that the following is helpful to you. 309 310I recently purchased a HP-7110i CD-RW, which is the U.S. only version of 311what you have. The HP 7100 and 7110 CD rewritables use the ATAPI 312standard. Originally, the drives were not supported under Linux (due to 313some inconsistencies with SCSI translations between the kernel and the 314CD), but that problem has just recently been fixed. There are some kernel 315and cdrecord patches that have been made to support this device that have 316yet to be officially incorporated into cdwrite and the kernel. In order to 317get your drive supported under Linux, you will have to do the following: 318 3191. Get the proper version of cdrecord. 320 321As of this writing, I am just getting ready to test Joerg's new cdrecord. 322I am currently operational on cdrecord-1.5, so I know that works, and I 323have attached patches for that version. 324 325If you are in a hurry, you can download ver. 1.5, apply patches, and 326rock-n-roll. You may want to wait, though. Up to you ;). The version with 327ATAPI support is cdrecord-1.6alpha5. I'm not sure if the current kernel 328patches are valid for this version, but i'll know soon enough. 329**** They are valid **** JS 330 331BTW, the new version of xcdroast now supports cdrecord - this version 332is in beta testing, too (currently uses cdrecord-1.5 but cdrecord-1.6a5 333should work with the actual xcdroast too). 334 3352. Upgrade to kernel version 2.0.31 336 337IDE/SCSI translation was first added in this kernel. Because your CD-RW is 338an ATAPI device, it will support SCSI command sets. The translation 339allows you to map the device as a SCSI generic device. This will allow 340cdrecord to recognize it as a SCSI device. 341**** 2.0.31 still needs patches, get 2.0.35 or later **** JS 342 3433. Get the patches and apply them 344 345Attached find kernel patches for kernel sources ide.h and ide-scsi.c, and 346cdrecord source scsi_cdr.c (version 1.5 only). 347**** Get cdrecord-1.6 or later **** JS 348 3493. Recompile kernel with SCSI emulation support 350 351If you do a "make menuconfig" or "make xconfig", select SCSI emulation 352under the category "Floppy, IDE, and other block devices". 353 354WARNING: 355Do not install SCSI support as a module - there is a bug in the makefile 356structure that will cause the compile to fail. Compile directly into the 357kernel. 358 3594. WARNING: Disable generic IDE/ATAPI CDROM support *** JS *** 360 361If you don't do this, the SCSI emulation will not work *** JS *** 362 3635. This is important too: 364You also need to enable SCSI and SCSI generic support *** JS *** 365 3666. Make sure that /dev/sg* exists. 367If they are missing, create them. 368 369Dave Cohen 370dcohen@richmond.infi.net 371(Patch instructions below) 372 373----------------------------------------------------------------- 374From: Danilo Fiorenzano <shade@juliet.gppsd.ab.ca> 375 376Anyway, here's what I did, using kernel version 2.0.33 I believe this 377is the proper way to get an HP-7100i to work (and as far as I can tell, 378any other IDE CD-writer unit): 379 3801) patch the kernel as described by README.ATAPI 381 3822) save your current kernel config to an alternate file, then run 383 "make mrproper" 384 3853) run 'make menuconfig' or 'make xconfig', then choose "load config 386 from alternate file" to restore the original configuration 387 3884) In "Floppy, IDE and other block devices", disable "IDE/ATAPI CD-ROM 389 support" and enable instead "scsi emulation" 390 3915) in "SCSI support" enable "SCSI support", "SCSI CD-ROM support" and 392 "SCSI generic support", everything directly in the kernel. 393 3946) compile, install kernel/modules, reboot. Now, if everything went 395 fine, your CDROM units should show up with a message like: 396 "hdb: HP CD-Writer+ 7100, ATAPI CDROM drive - enabling SCSI emulation" 397 3987) run "cdrecord -scanbus" to make sure cdrecord can see the unit and 399 talk to it. The end. 400 401Don't forget that now -all- of your CD drives are seen as -SCSI- units 402by all programs (/dev/scd0 etc.), so you might want to relink 403/dev/cdrom to the proper scd<n> in order to get xcdplay or whatever to 404work again. 405 406 407------------------------------------------------------------------------------- 408NOTE: 409 4101) Actual cdrecord releases support ATAPI 411 4122) Linux 2.0.35 or Linux 2.1.73 or later include ATAPI support 413------------------------------------------------------------------------------- 414From whampton@staffnet.com Fri Jan 14 05:21:34 2000 415From: "W. Wade, Hampton IV" <whampton@staffnet.com> 416 417You may wish to include/append these notes to your ATAPI notes.... 418 419I have my 4X Acer CD-R/RW ATAPI drive working with Linux. My platform 420is 421RedHat 6.1 with kernel 2.2.14. My first ATAPI CD device is a DVD with 422the second 423the CD-R. I made the following changes: 424 425Steps: 426 4271. Identify which device is the CD-R -- in my case the fourth ATAPI 428device, /dev/hdd. 429 4302. Compile the kernel to include ATAPI CDROM and SCSI emulation: 431 432 Under the block devices menu: 433 Y or M Include IDE/ATAPI CDROM support 434 Y or M SCSI emulation 435 4363. Build and install the upgraded kernel. 437 4384. If you selected modules, add them to the /etc/conf.modules file. 439 4405. In the /etc/lilo.conf file add an append line for ide-scsi, in my 441case: 442 append = "hdc=ide-scsi hdd=ide-scsi" 443 4446. Reboot to the new kernel and make sure the ide-scsi module is loaded 445 446 /sbin/lsmod | grep ide-scsi 447 4487. Make a link from the proper SCSI device to a symbolic, e.g., 449/dev/cdrom: 450 In my case the DVD is the first CD, hence appears as /dev/scd0 to 451scd7 452 (cat /proc/scsi/scsi to get a full list of devices -- the first 453CD-ROM will 454 appear as scd0, etc.) With the current ATAPI-SCSI module, each CD 455 456 device appears as 8 SCSI devices (different logical units). If 457you have 458 two devices, like I do, you may have to make a node for the second 459device. 460 In my case I had to make scd8: 461 462 cd /dev 463 mknod scd8 b 11 8 464 465 Then make links, in my case: 466 467 ln -s scd0 cdrom 468 ln -s scd8 cdr 469 470 Note, many CD-ROM player programs expect the audio CD drive to 471 be located at /dev/cdrom (xplaycd, etc.), hence this link is 472recommended. 473 474 If you try to use /dev/hdc (or wherever your CD or CD-R is) 475after loading 476 the ide-scsi module, you may not be able to mount CD's or play 477audio 478 discs -- you have to use the new SCSI names for the device. 479 4808. Fix your /etc/fstab file to mount the /dev/cdrom and /dev/cdr 481 482 483/*--------------------------------------------------------------------------*/ 484From: Eduard Bloch <edi@gmx.de> 485 486Situation: 487 Linux: Kernel 2.2.15 (Debian package kernel-image-2.2.15) 488 Distribution: Debian Potato (deep freeze), i386 489 Devices: one CDRW-Writer, one CDROM-drive, both ATAPI 490 4911. Become root, try "grep hd.: /var/log/kern.log" to find out where your 492 ATAPI-devices are connected to (hd?-names). 4932. Edit your boot configuration file, eg. /etc/lilo.conf if you use 494 lilo or the batch-file if you boot via loadlin. 4953. Find a line where you can append additional kernel parameters, eg. 496 "append=" in lilo.conf or the loadlin-line in the batch file. 4974. Append sth. like this: "hdb=ide-scsi hdc=ide-scsi max_scsi_luns=1" 498 The hdX-parameters defines devices that should be mapped to SCSI 499 latter. You may do it with non-writers too, since the emulation layer 500 is almost complete, or let them out so the devices will use their 501 native drivers. 5025. Save the file, reinstall the bootloader (ie. running "/sbin/lilo") 5036. Call "modconf", load "sg" and "ide-scsi" from the SCSI-section 5047. Reboot Debian, watch while booting, you should see a line like this 505 "Detected scsi CD-ROM sr0 at scsi0, channel 0, id 0, lun 0". 506 Your old ATAPI devices virtually don't exist any longer, use the 507 SCSI equivalents instead. 5088. Become root, setup devices: 509 cd /dev 510 MAKEDEV sg scd 511 ln -s scd0 cdrom # NOTE: or cdrw, first check which drive is here 512 ln -s scd1 cdrw # NOTE: see above, maybe cdrom 513 Check the new SCSI settings: 514 cdrecord -scanbus 515 Setup cdrecord's environment - edit /etc/default/cdrecord: 516 CDR_DEVICE=cdrw 517 cdrw=1,0,0 4 8m 518 cdrom=1,2,0 0 0m 519 Input the right values, the fields are described in the manpage 520 of cdrecord. Alternatively, you may use this values as 521 cdrecord-parameter or take a frontend with an own configuration 522 scheme, then you don't need to modify /etc/default/cdrecord. 5239. It's done! Insert a CD and try "cdrecord -v -toc" 524/*--------------------------------------------------------------------------*/ 525He had constant buffer underrun problems: 526 527From: "Trenton D. Adams" <trenton.adams@telusplanet.net> 528 529I enabled DMA, and 32-bit mode on the CD-Writer using "hdparm". 530This fixed the writing problem. 531 532/*--------------------------------------------------------------------------*/ 533From: "Mario Moder" <clay-man@freenet.de> 534----- 535TEAC CD-W54E 536 537I recently installed a TEAC CD-W54E (an ATAPI CD-RW-Recorder) and I had 538problems with buffer underruns and other errors when burning a CD (with 539Linux and Windows 2000). My system has an old ASUS P/I-P55T2P4 Pentium 540mainboard with Intel PCI-Bus-Master-IDE (I think the chipset is an Intel 541430HX and the IDE controller is an 82371SB). The harddisk is the master on 542the primary IDE channel, and the CD-Recorder is the master on the secondary 543IDE channel. 544 545After turning off DMA for the CD-Recorder AND the harddisk, the drive had no 546longer problems with burning a CD. You can try the following things to make 547it work, if you have similar problems with a similar hardware configuration: 548 549For Linux (Kernel 2.2.19): 550Turn off "Enable DMA by default" in the kernel (and then compile a new 551kernel), if you had it turned on or use "hdparm" to turn of DMA for both the 552CD-Recorder and the harddisk 553 554For Windows 2000: 555In the Device Manager go to "IDE ATA/ATAPI-Controller" and open the 556properties for the first and second IDE channel. There you change the mode 557of the devices from DMA to PIO. 558----- 559 560/*--------------------------------------------------------------------------*/ 561 562Hints for the Linux Packet code in ide-cdrom.c: 563 564 WARNING! It seems that this driver does not allow to send all 565 SCSI commands. A command that definitely fails is READ FULL TOC. 566 For this reason, you cannot read those 'defective' audio CDs 567 with broken TOC when you use this interface. 568 569 Thanks to Alexander Kern <alex.kern@gmx.de> for the idea and first 570 code fragments for supporting the CDROM_SEND_PACKET ioctl() from 571 the cdrom.c kernel driver. Please note that this interface in principle 572 is completely unneeded but the Linux kernel is just a cluster of 573 code and does not support planned orthogonal interface systems. 574 For this reason we need CDROM_SEND_PACKET in order to work around a 575 bug in the linux kernel that prevents to use PCATA drives because 576 the kernel panics if you try to put ide-scsi on top of the PCATA 577 driver. 578 579 The code is currently in "status nascendi" but usable with some trade offs. 580 581 To use: call e.g. 582 583 cdrecord -scanbus dev=ATAPI: 584 585 cdrecord -dao -v speed=24 dev=ATAPI:0,0 .... 586 587 Be careful! This code is only needed in order to be able to use 588 PCATA CD-writers on notebooks because there is a severe kernel bug. 589 Unfortunately, this bug causes the kernel to hang (and force you 590 to reboot) if you try to call: 591 592 cdrecord -scanbus 593 594 without the dev=ATAPI: option. 595 596 In this case cdrecord will hang infintely and unkillable 597 in open("/dev/sg1", 2) => you need to reboot :-( 598 599 Repeat by: Insert a PCATA CD-Writer in a Sony VAIO notebook and run 600 cdrecord -scanbus. 601 602